Analysis of Emotional and Behavioral Data of Students Experiencing Self-blaming and Exhaustion Using Decision Tree and Random Forest Methods (Case Study: SMKN 1 Sei Rampah)

Siti Mentari
Muhammad Irfan Syarif

Instal : Jurnal Komputer
Vol 18 No 02 (2026): Instal : Jurnal Komputer

Self-blaming and emotional exhaustion are key risk indicators for reduced engagement and well-being among students. This study applies interpretable (Decision Tree) and ensemble (Random Forest) classifiers to identify four risk groups: normal, self-blaming only, exhaustion only, and overlap (both). Using questionnaire-based emotional and behavioral indicators (Likert 1–5), the workflow follows CRISP-DM with preprocessing, stratified 80/20 split, class-imbalance handling, and evaluation using accuracy and macro/weighted F1. Results show low overlap prevalence but clinically meaningful high-risk subgroup. Random Forest achieves higher overall performance, while Decision Tree provides actionable rules for school counseling.